Scaffold Hopping Computational Approach for Searching Novel β-Lactamase Inhibitors

We present a novel computational ligand-based virtual screening approach with scaffold hopping capabilities for the identification of novel inhibitors of beta-lactamases which confer bacterial resistance to beta-lactam antibiotics. The structures of known beta-lactamase inhibitors were used as query ligands, and a virtual in silico screening a database of 8 million drug-like compounds was performed in order to select the ligands with similar shape and charge distribution. A set of numerical descriptors was used such as chirality, eigen spectrum of matrices of interatomic distances and connectivity together with higher order moment invariants that showed their efficiency in the field of pattern recognition but have not yet been employed in drug discovery. The developed scaffold-hopping approach was applied for the discovery of analogues of four allosteric inhibitors of serine beta-lactamases. After a virtual in silico screening, the effect of two selected ligands on the activity of TEM type beta-lactamase was studied experimentally. New non-beta-lactam inhibitors were found that showed more effective inhibition of beta-lactamases compared to query ligands.